ENERGY CONDITIONS AND SCALAR FIELD COSMOLOGY
by
SHAWN WESTMORELAND
B.S., University of Texas, Austin, 2001 M.A., University of Texas, Austin, 2004 Ph.D., Kansas State University, Manhattan, 2010
A REPORT
submitted in partial fulfillment of the requirements for the degree
MASTER OF SCIENCE
Department of Physics College of Arts and Sciences
KANSAS STATE UNIVERSITY Manhattan, Kansas 2013
Approved by:
Major Professor Bharat Ratra Copyright
Shawn Westmoreland
2013 Abstract
In this report, we discuss the four standard energy conditions of General Relativity (null, weak, dominant, and strong) and investigate their cosmological consequences. We note that these energy conditions can be compatible with cosmic acceleration provided that a repulsive cosmological constant exists and the acceleration stays within certain bounds. Scalar fields and dark energy, and their relationships to the energy conditions, are also discussed. Special attention is paid to the 1988 Ratra-Peebles scalar field model, which is notable in that it provides a physical self-consistent framework for the phenomenology of dark energy. Appendix B, which is part of joint-research with Anatoly Pavlov, Khaled Saaidi, and Bharat Ratra, reports on the existence of the Ratra-Peebles scalar field tracker solution in a curvature-dominated universe, and discusses the problem of investigating the evolution of long-wavelength inhomogeneities in this solution while taking into account the gravitational back-reaction (in the linear perturbative approximation). Table of Contents
List of Figures vi
Acknowledgements vii
Dedication viii
1 Energy conditions in classical General Relativity1 1.1 Introduction ...... 1 1.2 The weak energy condition ...... 5 1.3 The null energy condition ...... 8 1.4 The dominant energy condition ...... 12 1.5 The strong energy condition ...... 16 1.6 Convergence conditions ...... 18 1.7 The interrelationships between energy conditions ...... 21
2 Cosmology 24 2.1 Introduction ...... 24 2.2 How energy conditions constrain the cosmic fluid ...... 25 2.3 The constituents of the cosmic fluid ...... 28 2.3.1 Matter ...... 30 2.3.2 Radiation ...... 31 2.3.3 The scalar field ...... 31 2.4 Dark energy ...... 37 2.5 On the possible equations of state ...... 38 2.6 The effective cosmological fluid ...... 40 2.7 The Ratra-Peebles scalar field ...... 42 2.8 Conclusion ...... 44
A The Einstein-Hilbert action with boundary terms 50 A.1 Introduction ...... 50 A.2 The cosmological part ...... 51 A.3 The Ricci part ...... 52 A.4 The boundary term ...... 53 A.5 The matter part ...... 54 A.6 Conclusion ...... 54
iv B On the existence and stability of the Ratra-Peebles scalar field tracker solution in a curvature-dominated universe 56 B.1 Introduction ...... 57 B.2 A special solution ...... 59 B.3 The attractor property ...... 61 B.4 Gravitational back-reaction ...... 65
Bibliography 73
v List of Figures
1.1 Interrelationships between energy conditions and convergence conditions. Ac- cording to the conventions of the present work, the case Λ < 0 corresponds to a repulsive (de Sitter) cosmological constant and the case Λ > 0 corresponds to an attractive (anti-de Sitter) one...... 22
2.1 For the toy model of Example 2.3.1, we have plotted a(t) (solid blue), φ(t) (dot-dashed red), and w(t) (dashed orange) all on the same graph with time t running along the horizontal axis. We have set c1 = c2 = 0 and c3 = 1/3. . 36 2.2 A list of substances...... 39
vi Acknowledgments
I would like to express my thanks to Bharat Ratra, for his patient support as my research advisor in the Physics Department at Kansas State University. I wish to spotlight Anatoly Pavlov for very special thanks. The content of Appendix B is part of a joint work that he and I were involved with. I am thankful for his suggestions and useful discussions on certain aspects of this work. I would also like to send my warmest thanks to Larry Weaver, Andrew Ivanov, Omer Farooq, Jianxiong “Jason” Li, Data Mania, Raiya Ebini, Pi-Jung Chang, Louis Crane, Alan Guth, Andy Bennett, Khaled Saaidi, Amit Chakrabarti, Michael J. O’Shea, Mathematica, and especially my loving wife Lei Cao for her endlessly dependable support.
vii Dedication
To Omer: You expressed how eager you were to read this thesis. Here it is.
viii Chapter 1
Energy conditions in classical General Relativity
1.1 Introduction
The Einstein field equation describes the relationship between the stress-energy tensor Tµν of matter-fields and the geometrical properties of spacetime. In units where c = G = 1 (our preferred choice of units in the present work), the Einstein field equation reads: 1 R − Rg + Λg = 8πT , (1.1) µν 2 µν µν µν µν where gµν is the metric tensor, Rµν is the Ricci tensor, R := g Rµν, and Λ is the cosmological constant. In AppendixA, the Einstein field equation is derived using a variational technique.
In principle, one can take any metric gµν imaginable (for example, a traversable wormhole discussed below) and — as long as its second partial derivatives exist — plug that metric into the left hand side of (1.1) to produce the stress-energy tensor corresponding to that metric. In this way, exact solutions to (1.1) can easily be constructed, but the stress-energy tensors will not necessarily be physically reasonable. It is therefore useful to impose one or more energy conditions. Energy conditions serve to precisely codify certain ideas about what is physically reasonable. In the present work, we will study four energy conditions that are in standard usage. These are: the weak energy condition (WEC), the null energy condition (NEC), the dominant energy condition (DEC), and the strong energy condition (SEC).
1 Most of the source material for the present chapter comes from Section 4.3 of The Large Scale Structure of Space-time 1 by Hawking and Ellis, and Section 2.1 of A Relativist’s Toolkit 2 by Eric Poisson.
Example 1.1.1 (a traversable wormhole). The following metric, presented here as a line- element in spherical coordinates, describes a traversable wormhole (cf. Morris and Thorne3);
b(r)−1 ds2 = e2γ(r)dt2 − 1 − dr2 − r2 dθ2 + sin2 θdϕ2 , (1.2) r where b(r) and γ(r) are twice differentiable functions of r subject to certain restrictions:
(1) There is a finite positive radial coordinate r0 where b(r0) = r0. (This is the wormhole throat.)
(2)1 − b(r)/r ≥ 0 if r0 ≤ r < rh, where rh is positive and possibly infinite.
0 (3) b (r0) < 1.
(4) b(r) → −Λr3/3 as r → ∞.
(5) e2γ(r) → 1 + Λr2/3 as r → ∞.
Items (1) - (3) ensure that the spatial geometry has the shape of a spherically symmetric wormhole, with item (3) ensuring that the two spherical volumes on each side of the worm- hole throat are smoothly joined together (cf. Equation (54) in the paper by Morris and Thorne3). Items (4) and (5) ensure that the metric (1.2) corresponds to de Sitter, anti-de Sitter, or Minkowski spacetime in the asymptotic limit of large r depending on whether Λ < 0, Λ > 0, or Λ = 0, respectively. (According to the sign conventions of the present work, de Sitter spacetime has Λ < 0 and anti-de Sitter spacetime has Λ > 0.) When discussing the topic of traversable wormholes, one usually tries to limit the ac- celerations and tidal forces suffered by travelers who pass through the wormhole. We are here omitting such details but the interested reader is referred to the 1988 paper by Morris and Thorne.3 Morris-Thorne wormholes are asymptotically flat and horizonless, but since
2 we allow the possibility that Λ 6= 0, we should not insist on asymptotic flatness. Moreover, in the case where Λ < 0, one expects a cosmological horizon at r = rh. For a dedicated discussion on spherically symmetric traversable wormholes in de Sitter spacetime, the reader is referred to the 2003 paper by Lemos et al.4 According to the Einstein field equation (1.1), the wormhole (1.2) has a stress-energy tensor Tµν with non-vanishing components given by:
e2γ(r) (Λr2 + b0(r)) T = tt 8πr2 Λr3 + r + (2rγ0(r) + 1) (b(r) − r) T = rr 8πr2 (b(r) − r)) 2r2(r − b(r))γ00(r) + (rγ0(r) + 1) [2r(r − b(r))γ0(r) − rb0(r) + b(r)] − 2Λr3 T = θθ 16πr 2 Tϕϕ = Tθθ sin θ (1.3)
It is often convenient to express the stress-energy tensor in terms of an orthonormal basis
0ˆ 1ˆ 2ˆ 3ˆ at a base point p. In terms of an orthonormal basis (e0ˆ, e1ˆ, e2ˆ, e3ˆ), or its dual (e , e , e , e ), the metric tensor (locally) has the form of the Minkowski metric, and the stress-energy tensor has the form Tµˆνˆ where T0ˆ0ˆ is the energy density. For i, j ∈ {1, 2, 3}, Tˆiˆj = Tˆjˆi is the ˆi, ˆj-component of stress: the ˆi-component of the force exerted by matter-fields across a unit surface with normal vector eˆj. For each i ∈ {1, 2, 3}, Tˆiˆi is the normal stress in the (spacelike) eˆi direction (normal stress is pressure when T1ˆ1ˆ = T2ˆ2ˆ = T3ˆ3ˆ). For i 6= j ∈ {1, 2, 3}, Tˆiˆj is called shear stress. The components T0ˆˆi = Tˆi0ˆ, for i ∈ {1, 2, 3}, are (the negatives of) the components of energy flux as measured by an observer at p with 4-velocity e0ˆ (see Misner, Thorne, and Wheeler5 page 138).
Since Tµν is a symmetric second rank tensor, one can always find an orthonormal basis
(e0ˆ, e1ˆ, e2ˆ, e3ˆ) at each point, with e0ˆ future-directed, in which Tµˆνˆ has one of the following mathematically possible forms (cf. Hawking and Ellis1 pages 89 - 90, or Bona et al6 and the references therein):
3 ρ 0 0 0 0 p1 0 0 (first Segr`etype) 0 0 p2 0 0 0 0 p3 κ + ν 0 0 −ν 0 p1 0 0 , ν = ±1 (second Segr`etype) 0 0 p2 0 Tˆˆ Tˆˆ Tˆˆ Tˆˆ 00 01 02 03 −ν 0 0 ν − κ Tˆˆ Tˆˆ Tˆˆ Tˆˆ 10 11 12 13 = T2ˆ0ˆ T2ˆ1ˆ T2ˆ2ˆ T2ˆ3ˆ κ 0 −1 0 Tˆˆ Tˆˆ Tˆˆ Tˆˆ 30 31 32 33 0 p 0 0 (third Segr`etype) −1 0 −κ 1 0 0 1 −κ 0 0 0 ν 0 p1 0 0 , κ2 < 4ν2 (fourth Segr`etype) 0 0 p2 0 ν 0 0 −κ
The first Segr`etype corresponds to the unique case where the tangent space at each point has an orthonormal basis of eigenvectors of Tµν. As expressed in matrix form above, the
first Segr`etype has a timelike eigenvector e0ˆ with corresponding eigenvalue ρ, and for each i ∈ {1, 2, 3} there is a spacelike eigenvector eˆi with corresponding eigenvalue −pi. Note that the stress-energy tensor of a perfect fluid is of the first Segr`etype with p1 = p2 = p3
2 (cf. Poisson page 30). The second Segr`etype has two spacelike eigenvectors e1ˆ and e2ˆ corresponding to eigenvalues −p1 and −p2 respectively, and a double null eigenvector e0ˆ +e3ˆ corresponding to the double eigenvalue κ. Physically, a stress-energy tensor of the second
1 type describes massless radiation propagating in the direction e0ˆ + e3ˆ (Hawking and Ellis page 90). The third Segr`etype has a spacelike eigenvector e1ˆ corresponding to the eigenvalue
−p and a triple null eigenvector e0ˆ + e3ˆ corresponding to a triple eigenvalue κ. The fourth
Segr`etype has two real eigenvectors, e1ˆ and e2ˆ, which are both spacelike and correspond to 2 2 the eigenvalues −p1 and −p2 respectively. The condition κ < 4ν ensures that there are no other real eigenvalues. Hawking and Ellis1 wrote (page 90) that stress-energy tensors of the
4 third and fourth types do not arise in any known physical processes. In fact, by Theorems 1.3.1, 1.3.4, 1.4.1, and 1.5.1 in the present work, it follows that stress-energy tensors of the third and fourth types violate all four energy conditions. Returning to our wormhole example (1.2), the spherical coordinates (t, r, θ, ϕ) correspond to a coordinate basis (et, er, eθ, eϕ) where et = ∂t, er = ∂r, eθ = ∂θ, and eϕ = ∂ϕ.A transformation to an orthonormal dual basis can be reckoned by glancing at (1.2); one can use:
etˆ = eγ(r)et b(r)−1/2 erˆ = 1 − er r eθˆ = reθ
eϕˆ = r sin θeϕ (1.4)
Expressed in terms of (etˆ, erˆ, eθˆ, eϕˆ), the stress-energy tensor of our wormhole has the fol- lowing non-vanishing components:
Λr2 + b0(r) T = tˆtˆ 8πr2 (2rγ0(r) + 1) (r − b(r)) − Λr3 − r T = rˆrˆ 8πr3 2r2(r − b(r))γ00(r) + (rγ0(r) + 1) [2r(r − b(r))γ0(r) − rb0(r) + b(r)] − 2Λr3 T = T = θˆθˆ ϕˆϕˆ 16πr3 (1.5)
This is a tensor of the first Segr`etype.
1.2 The weak energy condition
µ ν µ The weak energy condition (WEC) asserts that Tµνv v ≥ 0 for all timelike vectors v (see Poisson2 page 30, or cf. Hawking and Ellis1 page 89). Since an observer with 4-velocity
µ µ ν v sees the local energy density as being Tµνv v , the WEC prohibits observers from seeing negative energy densities (Hawking and Ellis1 page 89, Poisson2 page 30). Although it may
5 seem reasonable to postulate that the WEC always holds, experiments have shown that it is violated by certain phenomena such as the Casimir effect. However, the current evidence suggests that there are strong limits on how severe such violations can be, globally (e.g., Poisson2 page 32).
µ ν Since the truth-value of the inequality Tµνv v ≥ 0 is unaffected if the nonzero vector vµ is replaced by any nonzero scalar multiple of vµ, it follows that the WEC is equivalent
µ ν µ to the statement that Tµνv v ≥ 0 for all normalized future-directed timelike vectors v . The wormhole spacetime of Example 1.1.1 does not satisfy the WEC. In fact, as we show in Section 1.3, it does not even satisfy the null energy condition, which is weaker than the WEC (see Theorem 1.3.1).
Theorem 1.2.1. If Tµˆνˆ is of the first Segr`etype, then the WEC is satisfied if and only if
1 2 ρ ≥ 0 and ρ + pi ≥ 0 for each i ∈ {1, 2, 3} (cf. Hawking and Ellis page 90, Poisson pages 30 - 31).
Proof. Let (e0ˆ, e1ˆ, e2ˆ, e3ˆ) be an orthonormal basis at a base point p, where e0ˆ is future- directed, and in which Tµˆνˆ explicitly has the diagonal form of the first Segr`etype. The set
FT p of all normalized future-directed timelike vectors at p is given by:
2 2 2 1/2 FT p = (1 + a + b + c ) e0ˆ + ae1ˆ + be2ˆ + ce3ˆ where a, b, c ∈ R (1.6)
µˆ 2 2 2 1/2 Let v eµˆ = (1 + a + b + c ) e0ˆ + ae1ˆ + be2ˆ + ce3ˆ be an arbitrary element of the set
FT p. Since Tµˆνˆ is of the first Segr`etype, we have:
µˆ νˆ 2 2 2 2 2 2 Tµˆνˆv v = (1 + a + b + c )ρ + a p1 + b p2 + c p3 (1.7)
The WEC requires that:
2 2 2 2 2 2 (1 + a + b + c )ρ + a p1 + b p2 + c p3 ≥ 0, for all a, b, c ∈ R (1.8)
In the particular case where a = b = c = 0, statement (1.8) implies that ρ ≥ 0. Taking
2 2 b = c = 0, statement (1.8) implies that (1 + a )ρ + a p1 ≥ 0 for all a ∈ R. Diving both sides 2 by a and taking the limit a → ∞ gives ρ + p1 ≥ 0. Similarly, ρ + p2 ≥ 0 and ρ + p3 ≥ 0.
6 µˆ 2 2 2 1/2 To prove the converse, let v eµˆ = (1 + a + b + c ) e0ˆ + ae1ˆ + be2ˆ + ce3ˆ be an arbitrary element of the set FT p and suppose that ρ ≥ 0 and ρ + pi ≥ 0 for each i ∈ {1, 2, 3}. One
2 2 2 2 2 2 has the following four inequalities: ρ ≥ 0, a ρ + a p1 ≥ 0, b ρ + b p2 ≥ 0, and c ρ + c p3 ≥ 0.
2 2 2 2 2 2 When added together, these give (1 + a + b + c )ρ + a p1 + b p2 + c p3 ≥ 0. That is,
µˆ νˆ µˆ Tµˆνˆv v ≥ 0 for any normalized future-directed timelike vector v .
Theorem 1.2.2. If Tµˆνˆ is of the second Segr`etype, then the WEC is satisfied only if κ ≥ 0,
ν = +1, and κ + pi + 1 ≥ 0 for each i ∈ {1, 2}.
Proof. Let (e0ˆ, e1ˆ, e2ˆ, e3ˆ) be an orthonormal basis at a base point p, where e0ˆ is future- directed, and in which Tµˆνˆ has the explicit form given by the second Segr`etype.
µˆ 2 2 2 1/2 Let v eµˆ = (1 + a + b + c ) e0ˆ + ae1ˆ + be2ˆ + ce3ˆ be an arbitrary element of the set
FT p. Since Tµˆνˆ is of the second Segr`etype, the WEC requires, for all a, b, c ∈ R:
2 2 2 2 2 2 1/2 2 2 (1 + a + b )(κ + ν) + 2c ν − 2νc(1 + a + b + c ) + a p1 + b p2 ≥ 0 (1.9)